Political constraints to growth in an economic union

Journal of Public Economics 93 (2009) 989–997

Contents lists available at ScienceDirect

Journal of Public Economics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p u b e

Political constraints to growth in an economic union☆ Michele Ruta Economic Research Division, World Trade Organization, Rue de Lausanne 154, 1211 Geneva, Switzerland

a r t i c l e

i n f o

Article history: Received 18 March 2008 Received in revised form 3 February 2009 Accepted 26 February 2009 Available online 17 March 2009 JEL classification: D72 F42 O30 O40 Keywords: Economic unions (De)centralization Lobbying Economic growth

a b s t r a c t This paper studies the political economy of growth in an economic union such as the EU. In the spirit of Acemoglu, Aghion and Zilibotti [Acemoglu, D., Aghion, P. and Zilibotti, F., 2006a, Distance to frontier, selection and economic growth, Journal of the European Economic Association, 4:1, 37–74; Acemoglu, D., Aghion, P., and Zilibotti, F., 2006b, Growth, development and appropriate versus inappropriate institutions, mimeo MIT.], as the economy approaches the world technology frontier, structural reforms that increase competition in intermediate goods sectors are necessary to boost innovation and productivity growth. Reforms, however, raise the opposition of incumbents and, therefore, are politically difficult to implement. When there are important cross-border policy spillover effects, national governments are more easily captured by vested interests, as they fail to internalize the benefits of reforms on the rest of the union. In this situation, productivity growth may be sluggish and the economy can fail to converge to the frontier. On the other hand, when policy is chosen by a union government (or a collective body that takes into account union welfare), the internalization of spillovers raises the perceived benefit of reforms and, consequently, lowers the ability of lobbies to obtain high levels of protection. © 2009 Elsevier B.V. All rights reserved.

1. Introduction In recent years, constitutional and economic reforms have been at the center of the political and academic debate in the European Union. A long list of economic analyses have stressed the importance of structural reforms in product and labor markets to unleash the innovation and growth potential of the EU.1 The academic debate, however, generally fails to acknowledge the relationship between constitutional (e.g. the allocation of competencies between the EU and national governments) and economic reforms; in short, the link between the political and economic future of the EU. This paper investigates the nature of political constraints to growth in an economic union under different constitutional (or political) regimes. Let me review in a nutshell the argument for economic reforms in the EU (which also provides the main motivation for the growth model employed in the paper). This argument is clearly described in the

☆ This article was written while I was a Fellow in the Economics Department of the European University Institute. I would like to thank the MacMillan Center at Yale University for hospitality during part of this project and Daron Acemoglu, Robin Boadway (the Editor), Daniel Brou, Thiess Buettner, two anonymous referees, and seminar participants at the 2007 Venice Summer Institute on Reinventing Europe for comments. Funding from the Centro Studi sul Federalismo at Fondazione Collegio Carlo Alberto is gratefully acknowledged. Errors are only my responsibility. Disclaimer: The opinions expressed in this paper should be attributed to the author. They are not meant to represent the positions or opinions of the WTO and its members and are without prejudice to members' rights and obligations under the WTO. E-mail address: [email protected]. 1 See, for instance, OECD (2007). 0047-2727/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jpubeco.2009.02.012

report to the European Commission known as the Sapir Report (2004). Growth in the post-war era in Europe was based on a set of economic institutions (among which non-competitive arrangements as various forms of state intervention in the economic activity) that favored investments.2 These institutions were suited at a time where the European economy was catching up with the technological leadership of the United States. However, once the technological gap narrowed, growth opportunities led by the adoption of existing technologies exhausted and Europe became more dependant on internally generated innovation. In the new economic environment, non-competitive arrangements that spurred investments in the postwar period are widely seen as imposing limits on innovation and growth. Hence, the need for reforms. As the Sapir Report puts it: “Europe's unsatisfactory growth performance during the last decades is a symptom of its failure to transform into an innovation-based economy” (page ii). The model builds on the work of Acemoglu, Aghion and Zilibotti (2006a) (henceforth, AAZ) who introduce the idea of appropriate (economic) institutions in an endogenous growth framework. They provide a model where certain rigid arrangements that reduce competition (e.g. high regulation) have positive effects on growth when an economy is far from the world technology frontier and the main economic problem is to fund investment in existing technologies. However, as the economy approaches the frontier, the potential

2 Eichengreen (2007) provides several examples on the role of non-competitive arrangements in boosting investment in Europe after World War II.

990

M. Ruta / Journal of Public Economics 93 (2009) 989–997

for growth by simply adopting existing technologies shrinks and anticompetitive arrangements are no longer optimal. Economic efficiency would require a change in economic institutions to more competitive relationships that favor innovation through a better selection of entrepreneurs and firms. However, governments might fail to implement such a change. The reason is that anti-competitive policies that favor growth through investment in early stages enrich incumbents. When economic power determines political power, governments find it difficult to reverse policies that are opposed by economically powerful constituencies. This political opposition to reforms ultimately reduces economic growth, possibly to the point where the economy stops converging to the frontier. In this paper, I extend this framework to consider an economic union under two alternative constitutional regimes. Under the first regime (which I will refer to as political separation), national governments decide policy independently. In the second regime (defined as political integration), a union government chooses policy for the entire union. Groups that stand to lose from reforms (i.e. the reduction of anticompetitive regulations) lobby national governments under political separation and the union government under integration. The model shows that in an economic union approaching the world technology frontier (i.e. which would benefit from economic institutions that promote competition), anti-competitive regulation is higher under political separation than under political integration, if cross-border policy spillovers are large. The reason is that, due to the cross-border effect of the policy reform, national governments fully internalize the political cost of reducing regulation while only partially internalizing its benefit. This entails that incumbent firms find it easier to lobby for (and obtain) inefficiently high regulation when national governments act separately. This political economy distortion affects economic performance, implying that growth will be slower under political separation than under integration. Moreover, it is possible to have an equilibrium such that an economic union converges to the technology frontier under political integration, while it fails to converge under separation. This article is at the cross-road of three recent lines of theoretical research. First, is the literature on the political economy of fiscal federalism. In particular, Brou and Ruta (2006), Bordignon et al. (2008), Redoano (2003) and Lockwood (2007) look at the allocation of competencies between different levels of government (local and central) when special interest politics is taken into account. Second, the literature on political (dis)integration and growth as Alesina, Spolaore and Wacziarg (2000), Alesina et al. (2005), Spolaore and Wacziarg (2005) and Brou and Ruta (2007). These papers focus on the channels through which political integration and separation affect economic growth. Finally, this paper contributes to the recent literature on fiscal federalism and growth (Oates, 1993; Brueckner, 2006; Hatfield, 2007; Koethenbuerger and Lockwood, 2007), which studies how centralization and decentralization of policy-making affect economic growth. The rest of the paper is organized as follows. Section 2 describes the model, while Section 3 studies the political economy of reforms and growth in an economic union under political integration and separation. Concluding remarks follow.

2.1. Production A unique final good, y, is produced at time t in all countries of the union. This final good is produced competitively using labor and intermediate inputs according to the following aggregate production function:

yt =

m 1 1− α X 1− α α Lt Ait xit ; α i=1

ð1Þ

where Ait is the productivity of intermediate input i at time t, Lt and xit are respectively labor and the flow of input i used in final good production at time t and α ∈ (0,1). The final good is the numeraire in this economy (with a price normalized to 1) and is used in the production of intermediate goods. Condition (1) assumes the Armington (1969) technology, where intermediate goods are differentiated by origin, so that input i is supplied by country i. This technology can be rationalized with the presence of country-specific knowledge (human capital) in the production of input i. Each intermediate good is produced by a national monopolist that has access to the most productive technology Ait and then is sold to final good producers in the market independently of their location. Shares of national monopolists are indivisible and non tradable and are owned by a small fraction of individuals in each country. More importantly, since these countries have formed an economic union, there are no costs associated to trade. This will provide the channel through which the effects of policy in one country (to be introduced below) spill over onto other countries in the union.3 The intermediate good producer has access to a linear technology and transforms one unit of final good into one unit of intermediate good. The national monopolist faces a competitive fringe of imitators (possibly from other countries of the economic union) that can copy its technology and produce an identical intermediate good. Imitation of the production process of intermediate i in another country of the union is successful with probability χi ∈ (0,1). Several reasons may rationalize a positive probability of failure in foreign imitation. Namely, this may be due to the adjustment process required to employ country i-specific knowledge in a different country of the union. Importantly, this probability of failure in foreign imitation renders intermediate good markets in the economic union less than perfectly integrated. Competition in national intermediate good sectors is influenced by government regulation that limits entry of both foreign and national potential competitors. Because of this regulation, the competitive fringe faces higher costs of production and needs additional units of the final good at time t to produce one unit of the intermediate good i. Namely, the higher the level of government regulation in country i and in the rest of the union, and the less competitive will be the market for the intermediate good i. The existence of this fringe, however, forces the national monopolist to charge the limit price pit = ð1 − χ i Þnit + χ i nmin t ;

ð2Þ

2. A stylized model of growth in an economic union Consider an economic union with a population of size 1 of overlapping generations of two-period lived agents and formed of m countries indexed by i = 1,2,…,m. These countries have equal size (1 / m) and similar economic and political structures. Namely, each country is populated by the same number of capitalists, who own firms, and workers, who supply their labor inelastically. Agents in the union have identical linear preferences in the consumption of the only final good produced. This simple framework allows us to focus on the production side of the economy.

where ξit and ξmin t are respectively the amount of regulation in country i and the minimum level of anti-competitive regulation in the union at time t.4 3 One could argue that intermediate goods are mostly used as inputs by national firms. While this argument is certainly true, the idea captured by this production function is simply that aggregate productivity in an economic union is influenced by the productivity of the different intermediate sectors in member countries. 4 This limit price is an equilibrium under the assumption that (1 − χi)nit + χinmin t is not so high that the national monopolist prefers to set a lower price. This is insured by the assumption that (1 − χi)nit + χ i nmin t V α1 .

M. Ruta / Journal of Public Economics 93 (2009) 989–997

Condition (2) implies that the limit price for the monopolist in country i depends on the regulatory choice of the government in country i and on the probability of successfully imitating input i in the rest of the union. If a government in the union reduces its level of anticompetitive regulation at time t, it lowers the cost of production of all inputs–domestic and not–as all intermediate goods can be successfully imitated with positive probability by foreign competitors. Therefore, national monopolists have to reduce their price to maintain their business, up to the point where competitors would not find it convenient to imitate even in the country of the union with the minimum level of regulation. Final good producers take intermediate good prices and the wage rate wt as given and solve the following maximization problem ( ) m m max X 1 1− α X 1− α α Lt ; x1t ; :::; xmt yt − pit xit − wt Lt s:t: yt = Lt Ait xit : α i=1 i=1 From the first-order conditions, we obtain the demand of intermediate good xit from the final good sector: −

1

xit = Lt Ait pit 1 − α :

ð3Þ

Given the equilibrium price (Eq. 2) and the demand from the competitive final good sector (Eq. 3), equilibrium profits of the monopolist in country i are: πit = δit Ait ; where δit u Lt f½ð1 − χ i Þnit + χ i nmin t  − 1g½ð1 −χ i Þnit + χ i nmin t 

ð4Þ 1 −1− α

:

The function δit(ξit) is monotonically increasing in ξit in the relevant range, therefore–other things the same–a less competitive national market implies higher profits for the national monopolists. From Eqs. (1)–(3) and the first-order condition of the maximization problem for final good producers with respect to labor, we obtain the equilibrium wage rate and aggregate output in the economic union, respectively 8 9 h  i− α = X 1− α < −1 −α α 1−α wt = Lt Ait nit + Ajt 1 −χ j njt + χ j nmin t : ; α j ≠ i 8 9 h  i− α = X 1 < −1 −α α 1−α + Ajt 1− χ j njt + χ j nmin t yt = Lt Ait nit : ; α : j ≠ i

ð5Þ ð6Þ

As countries have equal size and economic structure, each one of them will produce an equal share of the final output given by yit = ymt . Anti-competitive regulation in country i will have two effects. First, it will reduce aggregate final output in the national as well as in the union economy because of standard monopoly distortions (from Ayt it condition (6) we have that An b0 and Ay Anit b0). Notice, however, that it Ayit Ayt 1 Ayt = b for m N 1. This simply captures the main externality in m Anit Anit Anit this model: if country i increases anti-competitive regulation in its intermediate sector, it increases profits of its monopolist (see condition (4)), but reduces its final output only for a fraction equal to m1 (where, therefore, m–the number of member states in the union– captures the entity of the externality). Second, if regulation in country i is sufficiently low (i.e. ξit = ξmin t), then by lowering barriers to entry government i reduces the equilibrium price of all inputs to an extent that depends on the probability of foreign imitation. This effect decreases monopoly distortions.

991

2.2. Technological progress and growth: a reduced-form AAZ model Rather than repeating the analysis of the AAZ model, this section presents a reduced-form model based on Acemoglu, Aghion and Zilibotti (2006b).5 Economic growth is driven by technological progress, that is, by increases in the aggregate level of technology in the economic union m At ≡ Σi = 1 Ait. Each national monopolist can increase its Ait by two complementary processes: (i) innovation–i.e. the discovery of new technologies; and (ii) imitation–i.e. the adoption of existing technologies from the world technological frontier. Firms in the intermediate sectors are owned by capitalists and run by workers (entrepreneurs). Firms' productivity is determined by entrepreneurial skills. There are two types of entrepreneurs: high-skill and low-skill. Entrepreneurial skills are initially unknown and are revealed after an agent works as an entrepreneur for the first time. Last, investment projects of firms can be financed either through the retained earnings of old entrepreneurs or by the capitalist who owns the firm. The key economic decision in the AAZ model is whether to retain an entrepreneur with low skills or to replace him with a new entrepreneur with (on average) higher skills. The benefit of replacing a low-skill entrepreneur is traded off with the cost of financing investment projects, whereby the earnings of the retained low skill entrepreneur can be used to finance investments. Higher retained earnings will mitigate underinvestment problems that can emerge from market imperfections and moral hazard, but at the expense of making low skilled entrepreneurs more attractive to firms. The decision to retain an entrepreneur at time t is denoted by Rt ∈ {0,1}, where Rt = 1 and Rt = 0 correspond to retention and termination respectively. Denote with A t̅ the world technology frontier (and assume that it grows at the constant rate g) and with at ≡ At / A t̅ (where at ∈ [0,1]) the inverse measure of the economic union's distance to the world technology frontier at time t. We assume the following law of motion for productivity in this economy: 8 > > > <

 1  η + γat − 1 if Rt = 1 1+g P at =   > 1 > > η + γat − 1 if Rt = 0 : 1+g P

ð7Þ

where η ̅ N η̲ N 0 and γ ̅ N γ̲ N 0. This condition captures the main dynamics of the AAZ model. If the economy is distant from the frontier (at − 1 low), productivity growth is mostly driven by adoption of existing technologies and, therefore, growth is higher under retention (Rt = 1). AAZ refer to this as an investment-based strategy, because the main benefit of retaining (low-skill) entrepreneurs is their ability to reinvest retained earnings, thus effectively increasing the investment rate of the economy. On the other hand, as the economy gets closer to the world technology frontier (at − 1 high), growth increasingly depends on innovation and on the skills of the entrepreneur, thus growth is higher under termination (Rt = 0). AAZ refer to this as an innovationbased strategy, since the benefit in terms of growth of removing lowskill entrepreneurs is due to the process of “creative destruction”. Eq. (7) is depicted in Fig. 1. Economic growth is higher under the investment-based strategy for at − 1 b â (i.e. the economy is sufficiently far from the technology

5 This reduced-form captures the essential elements of the AAZ framework and greatly simplifies the discussion. This will allow us to focus on the endogenous determination of the policy and its implications for the growth process (the key aspects of the present work) at the cost of some additional assumptions. Results, however, do not depend on the use of this simpler model.

992

M. Ruta / Journal of Public Economics 93 (2009) 989–997

frontier), where the schedule (R = 1) implies a higher at for a given at − 1. On the other hand, productivity growth is larger under an innovation-based strategy (R = 0) for at − 1 N â (i.e. the economy is sufficiently close to the frontier), where â is determined by the intersection of the two schedules (R = 0) and (R = 1) in Fig. 1 and is η − η P given by aˆ = . Therefore, an optimal growth sequence is one in γ − γ

P

which the economy starts with an investment-based strategy and later switches to innovation (this sequence is depicted with the bold segments in Fig. 2). Naturally, nothing guarantees that the optimal strategy is an equilibrium. AAZ show that the economy will switch from the investment to the innovation-based strategy at at − 1 = ã, where ã can be larger or smaller than â depending on institutional factors (e.g. organization of credit markets), underlying economic conditions (e.g. incentives of entrepreneurs) and government intervention (the level of regulation ξit). Importantly, AAZ prove that reducing competition (i.e. increasing ξit) will increase the cut-off level ã. Intuitively, when intermediate good markets are less competitive, profits are higher (Eq. (4)) and so are the retained earnings of entrepreneurs. This induces capitalists to retain old entrepreneurs a˜ ðnit Þ whatever their skills (i.e. A An N 0). Finally, notice that there is a it   level of ξit, call it ξtrap, such that a˜ ntrap = 1 + gη − γ uatrap , where P

the economy never shifts to the innovation-based strategy and does not converge to the frontier. Up to this point, I have considered economic policy ξit as exogenous. However, policies are endogenously determined through some political process. In particular, if economic power is related to political power through lobbying activities, capitalists (who benefit from anti-competitive policy) can influence the government's choice of restrictions to competition. In the next section, I model this endogenous policy determination in an economic union and discuss the effects on economic growth of two distinct constitutional regimes. 3. The political economy of growth in an economic union This section studies the policy formation process in an economic union at each point in time and its implications for economic growth. I consider two extreme situations: political separation and political integration. In the first scenario, national governments independently and non-cooperatively choose national regulations. In the second scenario, a union government is in charge of deciding the policy for the economic union. We proceed by comparing these two political economy equilibria. I assume that governments live for one period and are “politically motivated” as in the standard model of Bernheim and Whinston (1986) and Grossman and Helpman (1994). This is a political economy

Fig. 2. Optimal growth sequence.

game where politicians care about social welfare, but can be influenced by lobbying activities of organized special interest groups. More specifically, old capitalists–i.e. the owners of national monopolies– are politically organized and can use their resources to lobby the government. The rest of the population has a clear interest in lobbying politicians as well, but it faces the standard collective action problem (as in Olson, 1965). This lobbying game between the old capitalists and the government determines the equilibrium level of regulation in the economy for each period and has important implications for the dynamics of the economy. I first describe the lobbying game under political separation and integration, I will discuss the effects on economic growth configurations in the last subsection. 3.1. Policy choice under political separation The political game has three stages. At the first stage, each lobby representing the interests of old capitalists offers to its national government a political contribution. This contribution is a binding commitment of payment and is contingent on the level of national regulation (ξit) chosen by the government at time t. At the second stage, each national government observes the contribution schedule and chooses the national level of regulation to maximize its objective function, taking as given the level of regulation in the other countries (i.e. independently of the actions of the other governments of the union). At the last stage, profits are realized and campaign promises are honored. The objective function of national government i at time t is a weighted average of social welfare (of current and future generations) and political contributions. More formally, government i maximizes

i

Gt = yit +

∞ X

β

t + z

yit

+ z

+ bcit ;

ð8Þ

z=1

t+z where yit + Σ∞ yit + z is the present discounted value of the z=1 β output (consumption) stream, with a discount factor β ∈ [0,1], cit denotes the political contribution to government i and b N 0 is a constant that defines the extent of the “political bias” in the government objective function.6 The higher b, the larger is the government's predilection for contributions. This parameter can be interpreted as capturing the level of checks and balances in government activity or, more in general, the quality of institutions. As extensively discussed in Grossman and Helpman (2001), this specification can be rationalized in several distinct ways.

Fig. 1. Investment-based and innovation-based growth strategies.

6 Notice that, as agents live for two periods, old capitalists only have an interest in influencing the policy at time t. As a result, contributions will not be contingent on future policy. On the other hand, a government which is in office for one period, cares about the current level of contributions.

M. Ruta / Journal of Public Economics 93 (2009) 989–997

As standard in this literature, I focus on a specific equilibrium where contributions take the following form cit = max ½0; π it − kit ;

ð9Þ

where kit is a constant optimally chosen by the lobby and πit is the incumbent's profit, which is given by Eq. (4). These contributions are often referred to as (globally) “truthful” as the shape of the contribution schedule reflects the effect that the policy has on the payoff it ðnit Þ it ðnit Þ to the lobby (i.e. AcAn = AπAn , ∀ ξit). Under truthful contribution it it functions, the solution to the lobbying game corresponds to the solution of a planning problem consisting of a weighted average of social welfare and the incumbent's profits. To gain some intuition, it is useful to start by characterizing the solution to a game where the government is “myopic”, in the sense that it disregards future generations, i.e. β = 0. Solving this problem for the symmetric equilibrium and under the additional assumption that the probability of carrying a successful imitation abroad is equal for all goods (i.e. χi = χ ∈ (0,1) ∀i), we obtain an explicit solution for the (Nash) equilibrium level of national regulation: n x˜ ðβ = 0Þ =

 −1 1 1 + χ ðm− 1Þ +1 : α αmb

ð10Þ

For χ b 1 and b b ∞, the extent of restrictions to competition in national markets for intermediate goods depends on the importance of the externality, as captured by the parameter m. The larger the number of countries, the less each one of them internalizes the negative effect of national regulation on the rest of the union and the higher is the extent of anti-competitive restrictions in each member country (larger ξ̃n (β = 0)). This effect is influenced by the parameters b and χ. The externality is more important the larger is the political bias of national governments, as a larger bias (a higher b) will induce each government to weigh more the interests of old capitalists and set a higher regulation (∂ξñ (β = 0) / ∂b N 0). Similarly, a change in the probability of successful foreign imitation affects the governments' choice of anticompetitive regulation. Namely, in a symmetric equilibrium, an increase in χ leads to lower regulation (∂ξñ (β = 0) / ∂χ b 0). Intuitively, if the probability of successfully imitating foreign inputs increases, by reducing regulation in country i, the government in i would lose the contributions from its own monopolist, but would reduce the cost of imitation of all m goods and, hence, reduce the equilibrium prices of all of them, thus increasing domestic output. In this case, the incentive to reduce regulation is strengthen and, therefore, equilibrium regulation is lower. Now, consider the general case of a “non-myopic” government who also cares about the stream of future output, i.e. for β N 0. As the nonmyopic government understands the effect that anti-competitive regulation has on welfare through productivity growth, would it choose a higher or lower level of regulation compared to the myopic benchmark? I assume that the economy starts off with an investmentbased strategy and distinguish three cases that depend on the position of the threshold where the economy would switch to innovation relative to the welfare maximizing cut-off value â: ã[ξ̃n(β = 0)] N â, ã[ξñ̃ (β = 0)] b â and ã[ξñ (β = 0)] = â. In the first scenario, ã[ξñ̃ (β = 0)] N â, under the regulatory stance chosen by the myopic government, the economy switches to an innovation-based strategy late and would benefit from an earlier change in strategy (in Fig. 2 a jump from the schedule R = 1 to R = 0). The government realizes that by decreasing regulation it lowers the threshold below ã[ξñ (β = 0)] and increases future output and welfare to the expenses of current profits. If, starting from ξñ (β = 0), the government reduces infinitesimally the regulatory stance, this involves no first-order change in current profits (and consumption), while it generates first-order gains in productivity Ait and, therefore, in

993

the present discounted value of future output. As the benefit from a marginal decrease in regulation is increased for β N 0, while the cost in terms of foregone contributions from capitalists is unaltered, the non-myopic government chooses in equilibrium a lower level of regulation, i.e. ξ̃n(β N 0) b ξñ (β = 0).7 This implies that ã[ξñ (β N 0)] b ã[ξñ (β = 0)]. In the second scenario, ã[ξ̃n(β = 0)] b â, the economy would switch too soon out of the investment-based strategy (in Fig. 2, from R = 1 to R = 0). For a small increase in regulation above ξñ (β = 0), there is no first-order loss in current output (and increase in current profits), while there are first-order gains in productivity and in the present discounted value of future output. Thus, the nonmyopic government will choose a higher regulation relative to the myopic government which implies a higher threshold ã[ξ̃n(β N 0)] N ã[ξñ̃ (β = 0)]. Finally, for ã[ξñ (β = 0)] = â, the economy would not benefit from an increase or a decrease in regulation relative to the myopic benchmark. This implies that ã[ξñ (β N 0)] = ã[ξñ (β = 0)]. The analysis therefore establishes: Lemma 1. ã[ξ̃n(β = 0)] N ã[ξ̃n(β N 0)] when ã[ξ̃n(β = 0)] N â; ã[ξ̃n(β = 0)] b ã[ξñ (β N 0)] when ã[ξñ (β = 0)]b â; ã[ξñ (β = 0)] =ã[ξñ (β N 0)] when ã[ξñ (β = 0)] =â. 3.2. Policy choice under political integration Consider now the second institutional environment–political integration–where member countries of an economic union delegate the power to choose national regulation in the intermediate sector to a union (i.e. supranational) government. The union government can decide the level of regulation in each country. Its objective function is given by u

Gt = yt +

∞ X

β

t + z

yt

u

+ z

+b

z=1

m X

cit ;

ð11Þ

i=1

which is a weighted average of union social welfare yt = Σz∞= 1 βt + z yt + z – i.e. the present discounted value of union aggregate output–and political contributions by national monopolists. The parameter bu(N0) captures the political bias of the union government, where bu ≶ b. The three-stage game between the lobbies representing the interests of old capitalists and the government has the same structure as before, with two main differences. First, by definition, the union government fully internalizes the effect of national regulations on union aggregate output. Second, lobbies simultaneously and non cooperatively offer contributions to the union government. As before, we start from the analysis of a myopic government (β = 0) as a useful benchmark. We can determine an explicit form for the (truthful) equilibrium level of national regulation under symmetry. This is equal to  −1 1 1 n x˜ ðβ = 0Þ = + 1 : α αbu

ð12Þ

As under political separation, the extent of anti-competitive regulation crucially depends on the political bias of the government (bu) in an intuitive way. Notice that the probability of successful foreign imitation, χ, and the size of the union, m, play no role in the determination of the equilibrium regulation under political integration, as the union government fully internalizes the spillover effect of the policy. Consider next the general case where β N 0. The non-myopic government will choose regulation at time t taking into account the 7 Naturally, this effect is stronger the larger the discount factor, β, in the government objective function.

994

M. Ruta / Journal of Public Economics 93 (2009) 989–997

effect that its choice has on the threshold ã at which the economy will switch from an investment-based strategy to an innovationbased strategy. The discussion follows the same logic of the political separation case and we can distinguish three different scenarios: ã[ξũ (β = 0)] N â, ã[ξũ (β = 0)] b â and ã[ξũ (β = 0)] = â. As union social welfare (i.e. the present discounted value of future aggregate output) is equal to national welfare multiplied by the constant m, the proof of Section 3.1 directly applies to this institutional environment as well. In sum, we have Lemma 2. ã[ξũ (β = 0)] N ã[ξũ (β N 0)] when ã[ξũ (β = 0)] N â; ã[ξ̃u(β = 0)] b ã[ ξ̃u(β N 0)] when ã[ ξ̃u(β = 0)] b â; ã[ ξ̃u(β = 0)] = ã[ξ̃u(β N 0)] when ã[ξũ (β = 0)] = â.

3.3. Regulation and growth under political integration and separation I proceed by making comparisons between the two political economy equilibria of the previous subsections and drawing implications for the dynamic pattern of the economy.

3.3.1. The politics of regulation I begin the discussion with the analysis of the case of myopic governments (with β = 0). This case has explicit solutions and helps convening the main intuition of the result (see the discussion below). From conditions (10) and (12), we find that there is a critical value of the spillover (i.e. the size of the union, m), such that anti-competitive regulation in an economic union is lower under political integration n than under separation (ξũ (β =  0) b ξ̃ (β =u0)) if and only if the spillχ Þb ˜ . This implies that the over exceeds this critical level m N ðb1 − um − χbu economy under political integration always switches to an innovationbased strategy at an earlier stage relative to political separation (i.e. ã[ξñ (β = 0)] N ã[ξũ (β = 0)]). Next, consider non-myopic governments, for which β N 0. As shown in the previous sections (Lemmas 1 and 2), in the general case the equilibrium regulation depends on the position of the cut-off value under the myopic benchmark relative to optimal threshold â, as the government realizes that its policy choice at time t will affect the growth pattern of the economy. A multiplicity of cases can emerge in equilibrium which depends on parameter values. Here, I focus on the scenario where spillovers are large, i.e. m N m̃.8 First, I study the case where the myopic government switches away from the investment based strategy too late both under political integration and separation (i.e. ã[ξñ (β = 0)] N ã[ξũ (β = 0)] N â). From Sections 3.1 and 3.2, we know that the non-myopic governments will choose in this case a policy that sustains an earlier equilibrium switch to the innovation-based strategy under both political regimes, i.e. ã[ξñ (β = 0)] N ã[ξñ (β N 0)] and ã[ξũ (β = 0)] N ã[ξũ (β N 0)]. Moreover, notice that the first-order effect of an infinitesimal reduction in regulation under political integration is larger than under separation, as the union government takes into account the effect of a reduction in regulation on the present discounted value of the output of the entire union, thus proving that ξñ (β N 0) N ξũ (β N 0). This, together with Lemmas 1 and 2, implies that ã[ξñ (β ≥ 0)] N ã[ξ̃u(β ≥ 0)] N â. Second, consider the opposite case in which the government switches too soon to the innovation-based strategy for β = 0 (i.e. ã[ξũ (β = 0)] b ã[ξñ (β = 0)] b â). The same arguments used above apply. The non-myopic governments will choose a policy that implies a later equilibrium switch to the innovation-based strategy (Lemmas 1 and 2). This incentive is greater for the union government relative to national governments. In this case, the equilibrium threshold under political separation can be to the right or to the left of the 8 The reader will have no difficulties in understanding the logic of the model in the opposite case (i.e. for m ≤ m̃). I also abstract from the special case where ã[ξñ (β = 0)] = ã[ξ̃u(β = 0)] = â.

threshold under integration, i.e. ã[ξ̃u(β N 0)] b ã[ξ̃n(β N 0)] b â. This discussion establishes: Proposition 3. Suppose that anti-competitive regulation is decided by a sequence of politically motivated governments andthat cross χ Þbu ˜ , then the border spillovers are sufficiently high m N ðb1 − um − χbu following two equilibrium configurations can emerge. In the first case, regulation is lower under political integration compared to separation, which sustains a lower cut-off level: ã[ξñ (β ≥ 0)] N ã[ξũ (β ≥ 0)] N â. In the second case, regulation can be higher or lower under political integration relative to separation, so is the cut-off level: ã[ξũ (β ≥ 0)]b ã[ξñ (β ≥ 0)] b â. When comparing the lobbying equilibrium under political separation and under integration there are two mechanisms at work, one operating through the internalization of the policy externality, the second through the change in the quality of political institutions. It is convenient to address them separately and focus first on the case where the economy jumps to the innovation strategy too close to the frontier with myopic governments (ã N â). First, assume that institutional quality does not vary under the two political regimes (i.e. bu = b). Direct observation of the symmetric equilibrium for the myopic government case (conditions (10) and (12)) shows that ξu(β = 0) b ξn(β = 0) for bu = b and m N 1. The intuition of this result is as follows: the non-cooperative structure of decision-making under political separation makes it easier for special interests to capture national governments and induce the adoption of stricter anti-competitive regulation compared to political integration. Recall that in the lobbying model based on the common-agency approach, contributions compensate the government for not choosing the socially optimal policy (Grossman and Helpman, 2001). The internalization of the externality under union increases the political cost of high regulation (the output loss of an excessive anti-competitive regulation) and induces lobbies to moderate their demands. If the government is non-myopic, β N 0, this effect is enhanced by the dynamic losses associated to excessively high regulation (and the more so, the larger the discount factor β). This explains why, in this case, structural reforms in an economic union–defined here as the reduction of national regulation that creates barriers to entry in the intermediate sector–can be more easily achieved under political integration than under political separation.9 Second, the assumption that the checks and balances on government activity, as captured by the parameters b and bu, are the same under political integration and separation plays a key role. However, the direction of change is not a priori clear. If institutional quality were higher at the supranational level (i.e. bu b b), then a myopic union policy maker would unambiguously choose lower national regulations. This would reflect the combination of the two effects: internalization of the externality and weaker influence of special interests on politicians. On the other hand, if the political bias were to be larger under political integration (i.e. a union government is more easily captured by special interests), then the benefit of political integration coming from the internalization of the externality would be weakened bythe increase in  the effectiveness of lobbying. Finally, the threshold of χ Þbu ˜ ðb1 − m mu for the myopic government highlights the role of the u − χb probability of successful foreign imitation, χ. Larger values of χ imply that the price of foreign inputs are more responsive to a decrease in domestic regulation. This mitigates the political economy mechanism 9 The extent of anti-competitive regulation in intermediate sectors (e.g. gas, postal services, rail transport, professional services, finance, electricity, telecommunications) in the US and the EU is consistent with these findings. According to Faini et al. (2006) and OECD (2007), between 1975 and 2003 regulation in these sectors has been consistently higher in the EU, where national governments have retained most of the decision power. Notice that this is not true for most final sector products (for which the regulatory stance is similar on the two sides of the Atlantic), where regulatory power has been centralized in Europe under the single market program of the 1990s.

M. Ruta / Journal of Public Economics 93 (2009) 989–997

Fig. 3. Growth sequence under political separation: the convergence case.

analyzed in this paper, as each government would internalize this price effect and, therefore, attach a greater value to a decrease in its regulatory stance.10 The finding of this section has a flavor similar to the well known Decentralization Theorem (Oates, 1972), which states that in presence of sufficiently large cross-border spillovers policy centralization dominates decentralization. The result is also reminiscent of the literature on strategic trade policy (Brander and Spencer, 1983, 1985), where the unilateral decisions of governments in the attempt to shift profits towards domestic firms is inefficient from the point of view of aggregate welfare. However, it should be stressed that in the present model the nature of the distortion is purely political, whereas the traditional fiscal federalism approach and the literature on strategic trade policy abstract from political economy considerations and correspond to the special case of this model where bu = b = 0 (i.e. benevolent governments). In this situation, it is immediate to see that there is no lobbying both under political integration and separation and that governments correctly understand the effect of their policy choice on welfare independently of the political regime. Finally, if the economy switches out of the investment-based strategy when it is far from the technology frontier (ã b â), the result is less clear-cut. Static considerations (the internalization of the policy spillover) would lead the union government to choose lower regulation (ξ̃u(β = 0) b ξñ (β = 0)). However, non-myopic politicians understand the dynamic effects of anti-competitive regulations and adjust the policy accordingly. The model, however, cannot specify which of these two effects dominates. 3.3.2. Growth dynamics The last step of this analysis consists in studying the effects that the political economy distortions have on the dynamics of the economy. Four interesting configurations describe the dynamic adjustment of this economy:

995

to retain old entrepreneurs (and maintain the investment-based strategy for a N â). Importantly, the economy fails to achieve the maximum growth rate for a range of values of a that depends on the political structure of the economic union. More precisely, political integration entails a lower level of anti-competitive regulation, which implies that the economy will switch to the innovation-based strategy closer to the growth maximizing strategy compared to political separation. For values of a ∈[ãu, ãn], growth is higher under political integration than under separation. In the long run, however, the economy switches to the innovation-based strategy independently of its political regime. • â b ãu b atrap b ãn. Political integration in the economic union leads to convergence to the technology frontier, while political separation does not (see Fig. 4). More precisely, for a b â there is growth with anti-competitive policies under both political regimes. For a N â, growth is higher if the economic union switches from an investmentbased to an innovation-based growth strategy. Differently from the previous configuration, the political regime here determines whether the economy will be stock in a non-convergence trap or will ultimately converge. In particular, under political integration regulation is lower and the economic union converges to the frontier; this will not be the case under political separation, where national monopolists can always induce governments to adopt higher anticompetitive policies. As in the previous case, the lack of coordination of national governments under political separation (and their failure to realize the negative effects of their policy on other member countries) is the reason why lobbying is more effective. However, in this case protection of incumbent firms not only retards growth temporarily, but pushes the economic union in a non-convergence trap. This equilibrium describes the most dangerous scenario for an economic union which is unable to coordinate policy choices. • â b atrap b ãu b ãn. The economic union does not converge to the frontier under neither political integration nor separation. The economy will converge for a b atrap, at which point it will stop converging to the technology frontier. Encouraging investments by reducing competition in the intermediate sector is initially optimal, however it will make it politically impossible for the economy to undertake a structural transformation. Lobbies representing the interests of old capitalists are able to induce the government to adopt an excessively high level of regulation regardless of the political regime (i.e. independently on what level of government–national or supranational–is in charge of deciding the policy). In this case, avoiding a non-convergence trap requires an improvement in the quality of checks and balances on government activity (i.e. a reduction of the political biases, bu and/or b), to induce governments–union or national–to be more responsive to the needs of the general public. Finally, notice that non-convergence traps are less likely to emerge under both political regimes, the larger the discount factor in the government objective functions (β).

• â b ãu b ãn b atrap.11 The economic union will eventually converge to the technology frontier independently of its political regime. This growth sequence is depicted in Fig. 3 (where the bold lines, as in the following figure, depict the equilibrium sequence under political separation). The economy starts with the investment-based strategy (R = 1), which is optimal for a b â. As the economy gets closer to the frontier (a N â), it would be optimal to enhance entrepreneurial selection i.e. to switch to (R = 0). However, lobbying by national monopolists induces governments to choose high levels of anticompetitive regulation, thus increasing the (short-run) convenience 10 For χ → 1, intermediate good markets in the economic union are fully integrated. In this case, the externality ceases to play any role and equilibrium regulation is the same under both political regimes. I thank an anonymous referee for bringing this point to my attention. 11 To simplify the notation, we have that ãn ≡ ã[ξñ (β ≥ 0)] and ãu ≡ ã[ξũ (β ≥ 0)].

Fig. 4. Growth sequence under political separation: the non-convergence case.

996

M. Ruta / Journal of Public Economics 93 (2009) 989–997

• ãu b ãn b â b atrap and/or ãn b ãu b â b atrap. The economic union abandons the investment-based strategy too soon. This equilibrium configuration requires some explanation. Recall that for ã b â the economy can reach a higher growth rate through government intervention, which offsets the underinvestment problem. As discussed in the previous section, government intervention might not stop when ã = â. These are the equilibrium sequences discussed in the three cases above. Now the question is the following: is it possible that political integration might induce the economic union to switch to an innovation-based strategy too soon (i.e. for ã b â)? If this could be the case, political integration would reduce growth for economic unions far from the frontier for values of a ∈ [ãu, â]. In the long run, the economy would converge to the technology frontier independently of its political structure. This suggests that there might be appropriate political (in addition to economic) institutions: i.e. optimal political organizations (as, for instance, the allocation of prerogatives between different levels of government in an international union) may differ with the stage of development. An issue clearly well beyond the scope of the present work. 4. Conclusions This paper presents a formal investigation of the political constraints to growth in an economic union. The model builds on the work of Acemoglu, Aghion and Zilibotti (2006a), who show that an economy that approaches the technology frontier might fail (or delay) to switch to a growth-enhancing innovation-based strategy. I study under what conditions this problem is more severe–and growth is lower–in an economic union. The model shows that, anti-competitive regulations can be lower under political integration compared to political separation when cross-border spillovers are sufficiently high. The reason is that union policy makers fully internalize the spillover effect of reforms (a reduction of regulation) and, as a result, this makes lobbying less effective for vested interests. In this case, the dynamics of the model has two interesting political economy equilibria. In the first one, the economy is able to sustain higher growth rates under political integration compared to separation, but eventually converges to the world technology frontier under both political regimes. In the second equilibrium configuration, the economic union is only able to reform and switch to an innovation-based strategy under political integration and is stock in a non-convergence trap under political separation. More generally, the ability of the union to converge to the technology frontier may depend on its constitutional structure. According to this model, a possible interpretation of the scarce success of the Lisbon Agenda (the series of objectives set in the European Council held in Lisbon in March 2000) is the inability of European countries to coordinate reforms in presence of relevant policy spillovers. These findings contribute to the literature in three ways. First, relative to the literature on political integration and growth (in particular Alesina et al., 2000), this paper finds that political integration and separation have a novel effect on growth. This effect results from the inability to implement growth-enhancing structural reforms in an economically interdependent world under full political separation. Second, the paper contributes to the literature on fiscal federalism. In a static sense, this paper confirms the importance of policy spillovers across different jurisdictions in the cost–benefit analysis of centralization when policy makers are influenced by lobby groups rather than being social welfare maximizers.12 In a dynamic sense, and contrary to most theoretical studies, the model finds conditions under which centralization can be beneficial to growth. As the empirical evidence on fiscal federalism and growth provides quite mixed results, a possible interpretation suggested by these findings is that it is the 12

As we discuss above, however, the logic of this result is different from the standard Decentralization Theorem.

appropriate allocation of competencies between different levels of government (rather than simply more or less decentralization) that may ultimately deliver better economic performance. The present work leans on some simplifying assumptions, because my prime concern was to sketch out the relationship between the allocation of competencies between different levels of government in an economic union and its growth performance. The need for future developments is, therefore, obvious. Two extensions strike me as particularly relevant. First, the political process under political integration and separation (and, hence, the outcomes of such processes) may vary substantially. This model allows for different degrees of political biases across regimes, but does not formalize the details of decision making in a centralized and in a decentralized system.13 A second extension may involve the use of dynamic common-agency games to model the lobbying process. The theoretical literature (Bergemann and Välimäki, 2003) and its applications to growth models (namely, Bellettini and Ottaviano, 2005) assume that governments only care about special interests' contributions. This paper, to the other extreme, uses the Grossman and Helpman (1994) set up, which allows to generalize the objective function of the policy maker, but excludes from consideration dynamic lobbying. A dynamic lobbying game where governments weigh public utility per se would represent an important extension, beyond its implications in this context.

References Acemoglu, D., Aghion, P., Zilibotti, F., 2006a. Distance to frontier, selection and economic growth. Journal of the European Economic Association 4 (1), 37–74. Acemoglu, D., Aghion, P., Zilibotti, F., 2006b. Growth, development and appropriate versus inappropriate institutions, mimeo MIT. Alesina, A., Spolaore, E., Wacziarg, R., 2000. Economic integration and political disintegration. American Economic Review, December, 1276–1296. Alesina, A., Spolaore, E., Wacziarg, R., 2005. In: Aghion, Philippe, Durlauf, Steven (Eds.), Trade, growth and the size of countries. Handbook of Economic Growth, vol 1. North Holland, Amsterdam, pp. 1499–1542. Part 2, Chapter 23. Armington, P.S., 1969. A Theory of Demand for Products Distinguished by Place and Production. IMF Stuff Papers, vol XVI, pp. 159–178. Bellettini, G., Ottaviano, G., 2005. Special interests and technological change. Review of Economic Studies 72, 43–56. Bergemann, D., Välimäki, J., 2003. Dynamic common agency. Journal of Economic Theory 111 (1), 23–48. Bernheim, D.B., Whinston, M.D., 1986. Menu auctions, resource allocation, and economic influence. Quarterly Journal of Economics 101 (1), 1–31. Besley, T., Coate, S., 2003. Centralized versus decentralized provision of local public goods: a political economy analysis. Journal of Public Economics 87 (12), 2611–2637. Bordignon, M., Colombo, L., Galmarini, U., 2008. Fiscal federalism and lobbying. Journal of Public Economics 92 (12), 2288–2301. Brander, J.A., Spencer, B.J., 1983. International R&D rivalry and industrial strategy. Review of Economic Studies 50, 702–722. Brander, J.A., Spencer, B.J., 1985. Export subsidies and international market share rivalry. Journal of International Economics 18, 83–100. Brou, D., Ruta, M., 2006. Special interests and the gains from political integration. Economics and Politics 18 (2), 191–218. Brou, D., Ruta, M., 2007. Economic integration, political integration or both? mimeo Columbia University and EUI. Brueckner, J., 2006. Fiscal federalism and economic growth. Journal of Public Economics 90, 2107–2120. Eichengreen, B., 2007. The European Economy Since 1945. Princeton University Press, Princeton NJ. Faini, R., Haskel, J., Barba Navaretti, G., Scarpa, C., Wey, C., 2006. Contrasting Europe's decline: do product market reforms help? In: Boeri, T., Castanheira, M., Faini, R., Galasso, V. (Eds.), Structural Reforms Without Prejudices. Oxford University Press, Oxford UK. Grossman, G.M., Helpman, E., 1994. Protection for sale. American Economic Review 84 (4), 833–850. Grossman, G.M., Helpman, E., 2001. Special Interest Politics. MIT Press, Cambridge, MA. Hatfield, J., 2007. Federalism, taxation and economic growth, mimeo, GSB, Stanford University. Koethenbuerger, M., Lockwood, B., 2007. Does tax competition really promote growth? mimeo, University of Warwick. Lockwood, B., 2007. Voting, lobbying and the decentralization theorem, mimeo, University of Warwick. 13 See Besley and Coate (2003) for a political economy model that provides an explicit formalization of the decision-making under centralization and decentralization. Ruta (2003) also discusses these issues within the context of lobbying models.

M. Ruta / Journal of Public Economics 93 (2009) 989–997 Oates, W., 1972. Fiscal Federalism. Harcourt, Brace and Jovanovich, New York. Oates, W., 1993. Fiscal decentralization and economic development. National Tax Journal 46, 237–243. Olson, M., 1965. The Logic of Collective Action. Harvard University Press, Cambridge, MA. Organization for Economic Cooperation and Development, 2007. Going for Growth: Economic Policy Reforms. available at www.oecd.org/eco. Redoano, M., 2003. Does centralization affect the number and size of lobbies? Warwick Economic Research Papers, No 674.

997

Ruta, M., 2003. The Allocation of Competencies in an International Union: A Positive Analysis. ECB WP No. 220, April 2003. Sapir, A., 2004. An Agenda for a Growing Europe: The Sapir Report. Oxford University Press, Oxford, UK. Spolaore, E., Wacziarg, R., 2005. Borders and growth. Journal of Economic Growth 10 (4), 331–386.